This type of acoustic signal generator is disclosed in, for example, Japanese Laid-Open patent application No. 59-26789. When one sound is produced, its tone generally changes delicately as time elapses. Specifically, since high frequency components contained in the sound attenuate readily whereas low frequency components attenuate to a small extent, the condition (tone) of the sound changes delicately as time elapses. Therefore, according to the system disclosed in the foregoing application, three kinds of acoustic data are preset which correspond to the waveform of tone of a sound when it has just been produced, the waveform of tone when a certain time has elapsed after sound production so that the sound has attenuated correspondingly, and the waveform of tone when another certain time has elapsed so that the sound has correspondingly become more attenuated, these acoustic data are previously stored in three storage circuits (ROM) respectively. Then the waveforms of three sounds to be produced in succession are modified by the use of the three kinds of acoustic data, the resulting sound outputs are added digitally together, and the resulting output is converted from digital to analog form and modified by the use of one envelope signal; as a result, the three sounds are combined.
In the foregoing system of combining three sounds by the use of one envelope signal, when the time constant of the envelope signal is large, attenuation is dull; thus, the strength of sound increases as the number of overlapping sounds increases. On the contrary, when the time constant of the envelope signal is small, the sound is attenuated quickly; accordingly, two or more sounds produced earlier become too weak to hear, making it difficult to discriminate the overlapping of sounds.